Your Personal Social Media Content Generator

This is a speculative design project where we explored the alternate presents and possible futures of human and technology. We generated the concept of Amanda, a purposefully-flawed digital personal assistant, to create a commentary towards the current interaction between people, AI, and social media.


Design Lead

Team Members

Aidee Cantu
Mane Chakarian
Taylor Wong


The development of AI and the increasing amount of its successful implementations in the field of social media has already started to make changes to people’s lives. As we begin to envision more about the future, it is important to consider cases where AI might not be effective. Specifically, what are the problems currently existing that would not be solved as a result of new interactions between human and AI?

A Commentary

Our solution, Amanda, serves as a commentary through scenarios we created for it. Amanda is designed as a personal social media assistant that allows people to stay fully connected to social media, without needing to put in any effort. The scenarios we created for Amanda transitions from probable and appealing to improbable and creepy. This process allows our audience to draw attention to and reflect upon their own social media habits and understandings towards AI.

Amanda is designed to function and progress through three phases.


It will gain access to user’s data, learn about their habits, and give them suggestions based on those habits.


Amanda will generate content for the user to choose from to post on their social media platforms.


Amanda will take full control over the user’s social media accounts and all of the content generations.


This project was exhibited after its completion and was appreciated by visitors, students, and faculties. The commentary was considered really clever, and it instantly triggered multiple discussions around the topic at the spot.

This project was featured in the second issue of the Berkeley Arts + Design initiative publication.

What are we looking at here

To understand how we came up with the concept of Amanda, we must journey back to the very beginning of our process. We started our journey by first looking into these three categories below.

Artificial Intelligence
Artificial Intelligence

The development of artificial intelligence has been heated discussed in recent years. Applications such as self-driving car attracted a lot of attention from the public. Many people wonder how AI technology will be able to revolutionize other fields as well.

UI design
Social Media

Social media has become a crucial part of most people’s everyday lives, as it provides means for people to connect, generate content and interchange ideas. Some of people’s activities start to revolve around social media, creating entirely new circumstances which evoke both opportunities and challenges.

UI design

As technology advances, many people develop more and more reliance on technology. Such reliance, while being beneficial on many occasions, inevitability creates problems. How to balance the benefit and issues became the core of the interaction between human and new technology.

As we conducted preliminary research into each of these areas, we looked into other speculative designs which focused on combining all the areas together. We were able to find a great amount of examples in films and tv series, where futuristic implementations of AI in social media played a huge part in shaping people’s lives. Some great examples would be the movie “Her” as well as the series Black Mirror.

Culture Probing

Now that we had a glance at what our problem space might look like, in order to envision the future, we decided to analyze the present first. Our first goal was to have a look at the existing relationship between social media and people. Due to the speculative nature of our project, Instead of starting interview sessions or sending out surveys, we decided to do something that would give us more natural and creative observations and provoke more inspirational responses - A culture probe.

What We Did - fakePhone

When we looked at the relationship between social media and people nowadays, we saw everyone connected all the time. People text daily, tweet every few hours, and scroll through Facebook whenever they get a chance.

Our assumption was that adding artificial intelligence into this relationship would inevitably reinforce it.

To help us create an artifact that would trigger more engagement from our potential participants, we reversed the assumption, and created a scenario much more interesting:

What if all technologies, not only AI, were to abruptly disappear? How would people react then? Would they still feel as connected to everyone they “communicated” with on a daily basis?

In search of answers for these questions, we created fakePhone. fakePhone is a notebook with the cover of a laser-cut phone, and with pages of “fake apps” that were designed to mimic tasks and functions of real apps. These fakePhones were then distributed to our participants, who were asked to replace their real phones with fakePhones for a few days.


From the fakePhones we collected back from our participants, we were able to generate a great number of insights which we then grouped and sorted using affinity mapping.

  • People share and edit the good parts of their life to showcase to others.

  • On social platforms, people are forced to put in a great amount of time and effort.

  • Social platforms become places where people compete with others for popularity.

  • Most people can’t live even a day without a phone.

  • Without specific instructions, some people find it hard to create content.

  • People need confirmation (likes) when they are creating creative content.

  • Online social platforms make it easy for everyone to express themselves in their own unique way.

  • People enjoy using platforms that are familiar with functionality.

  • People are used to interacting with certain apps in certain ways.

  • Lack of a framework can be a hindrance (too much freedom).

  • Some people rely on frameworks to help maximize creativity.

  • Frameworks do limit people's freedom of expression.

Intrinsic Value
Intrinsic Value
  • Creativity can seem like a chore sometimes.

  • People like to be creative but in limited quantities.

  • People are rarely self-motivated to produce creative content.

Key Insights

Aside from the insights above, we were able to discover three key insights that were most interesting to us.

People don’t want to spend that much time on social media.

People would give up freedom of expression for benefit brought by technology.

People appear to be more creative in a constrained environment or when following a fixed pattern.


We also identified the problems we are tackling to be:

People love to have creative content on their own social media pages...

...yet they are usually unwilling to actually do the hard work and generate those content.


Based on all the insights generated, and the problem we identified, we conducted several brainstorming sessions and came up with many potential design ideas. For these ideas, we were adding AI to the formula. We eventually narrowed down to four ideas, and for all of them, we created personas and scenarios to generate feedback.


We’re spending too much time updating our lives and too little time enjoying our own lives for ourselves.Amanda aims to solve this issue.

Powered by state-of-the-art machine learning, your virtual social media assistant is trained on your existing social media activity. Data collected from your current surroundings are then used by the assistant to automatically generate new content to be posted on your social media accounts.

No effort is required on your part: simply wear your Spectacles and go about your day, watching as all the likes and comments roll in!

Jason’s Story

27, Lawyer, NYC

While in school, he used to be able to keep up with his social media applications.

But now he simply hasn’t had time to post anything.

He recently logged into Facebook and saw that everyone’s so active and well-connected.

He wants to have that.

So he got Amanda, the new automated social content generator.

With Amanda, he got tons of followers, and reconnected with a lot of his old friends.

Jason is Happy.

True Feeling

Is it worth posting content that won’t garner many likes or positive comments? How do I know if what we’re about to post is really subpar content? Are my friends showering my posts with fake likes out of pity?

True Feeling removes any degree of uncertainty you might have by collecting the raw reactions of your friends. It utilizes your friends’ webcams to capture a snapshot of their face as they view your post. Using an automated facial expression recognition algorithm, True Feeling classifies their facial expression and compiles these results into a histogram that allows you to instantly recognize whether your recent post was total garbage.

Danny’s Story

23, Engineer, San Diego

Danny’s girlfriend is about to post something.

Before she publishes it online, she asks Danny for his opinion on the matter.

Not really paying attention, Danny says it's fine and so she posts it.

However, Danny happens to accidentally view her post first and his true feelings are exposed.

It turns out he doesn’t actually agree with what his girlfriend shared.

His girlfriend is upset that Danny lied.

Danny, seeing how upset she is, decides to simply be honest the next time she asks him something.

Danny is happy.


Today, there are many forms of communicating such as written communication, images, videos, audios, and gestures. And for the most part, this seems to work. However, there are situations where even a combination of all of these methods isn’t enough to get a point of view across.

That’s where Full-Feel comes in: Full-Feel helps capture your true emotions and allows you to accurately share your feelings with others.

Martha & Bobby's Story

Martha: 35, Neurosurgeon, SF
Bobby: 32, Teacher, SF

Martha and Bobby are married and have 3 kids.

Bobby generally stays at home and watches the kids.

Martha generally works long shifts and doesn’t get to see her family very often.

Last time she had a full conversation with Bobby was 2 weeks ago.

Bobby is feeling frustrated, upset, and lonely. He doesn’t feel like he’s being heard or understood.

So he decided to use Full-Feel.

Now Martha can really feel his pain, his loneliness and frustration.

She decided to spend more time with Bobby and the family.

Bobby and Martha are happy.


Friend-Less is a bionic implant for the human brain that allows one to effectively “delete” a friend from memory via a connected smartphone app.

Do you have a friend who just isn’t worth enough to you to justify a hit to your real-life follower-to-friend ratio? Simply open the Friend-Less app and remove that friend– you’ll literally forget that you ever knew him! Before you know it, you’ll be wondering how so many strangers seem to know you (i.e. how you have so many followers).

Stop fretting over your Twitter follower-to-friend ratios; your real-life perceived social status is what really matters!

Robert’s Story

24, Mechanic, SF

Robert was extremely popular in high school, but in college he found himself spread too thin across his social circles.

He knew plenty of people, but he found that many people who met him quickly forgot him.

Robert thought it was because the number of people he knew vastly exceeded the number of people that knew him.

Robert gets the Friend-Less implant and installs the accompanying app.

He purges the majority of his friends from his memory, thus obtaining a respectable follower-to-friend ratio.

Robert begins to respect himself more.

Robert is happy.

We picked Amanda, the personal social media content generator, as our final choice, since this idea resonated the best with our insights and targeted our problem in the most direct way.

Without AI
AI generate content for people

People don’t want to spend that much time on social media.

People don’t need to spend as much time on social media.

People would give up freedom of expression for benefit brought by technology.

People can follow directions generated by AI.

People appear to be more creative in a constrained environment or when following a fixed pattern.

People never need to do the work themselves.

People love to have creative content on their own social media pages...

People have content on their social media.

...yet they are usually unwilling to actually do the hard work and generate those content.

People can sit back and relax.

“What can go wrong?”

As an idea for speculative design, we envisioned Amanda to go beyond simply being a solution. We want to use Amanda as an example of “what can go wrong” by pushing it to its extreme to help people understand the absurdity of the problem we are facing. Amanda would be an over-engineered solution to the problem. She would allow us to spend zero time on social media, but at the same time remain fully-active participants on social media. Amanda will be something powerful without control, charming at first, but becoming more and more frightening as it evolves.

Who is it for

Based on our insights before, we decided on the base case for who should Amanda be built for: The overworked and overwhelmed – people who are too busy to keep their friends and followers updated on social media. Amanda is for people who want to create content yet unwilling to spend time or energy on that process.


To help the target users achieve their goals of remaining active online without actually being online, Amanda was created to have three important functions: Observing, Learning, and Generating. Amanda should first observe the user’s interaction and content generation patterns, and learn and improve at the same time to try and mimic user’s behaviors, and eventually generate all content for the user. We assigned the tasks into three phases that Amanda could gradually progress from one to another.

Three phases of Amanda

Once Amanda is installed and activated, it will automatically run in the background, and will go through a total of three phases.

During all phases
During all phases
  • Monitors user’s heart rate and body temperature for changes to detect emotions.

  • Record snippets of user’s day or take photos of the user’s surroundings when it detects new location or change of emotion.

  • Record when a user makes sounds or communicates with others.

Phase I  Learning
Phase I  Learning
  • Amanda Learns from the user: how comments are structured, images they respond to, when do they post, when do they comment, who they comment most to, etc.

  • Amanda begins to give advice about user’s posts.

  • Amanda will learn how users follow the advice given and adjust accordingly.

Phase II  Partial Automation
Phase II  Partial Automation
  • Amanda generates content for the user from learned data (generates text, captures Images, selects and apply filters, etc).

  • Amanda Asks user questions for better generations and further learning.

  • Amanda will strongly recommend what material for the user to post.

Phase III  Total Control
Phase III  Total Control
  • Amanda automatically posts comments and posts on the user’s behalf.

  • Amanda will decide when is the right moment and the right location to post things.

  • Amanda will select content from its collection of everyday recordings and footage.


After figuring out the general task of Amanda, we then looked into how would Amanda be able to fulfill them.

Hardware wise, Amanda requires a smartphone with internet, GPS and Bluetooth as well as a wireless earpiece, which is built in with a heartbeat sensor, temperature sensor, microphone, and camera.

Software wise,  Amanda should be an AI built into an app, which can run across most major OS platforms.

Interaction flow

We created the interaction flow to demonstrate how Amanda will interact with the users in different scenarios during different phases.


The onboarding process for Amanda is simple. All the user has to do is to click through next buttons like they would do for any other applications.

Phase I  Learning

During this phase, Amanda will show up when the user decides to post content for social media. Right now, Amanda will only give out suggestions on what to post, and will not interfere with the user’s normal behavior.

Phase II  Partial Automation

At this phase, Amanda will make the process of posting much easier by auto-generating options for the user to choose from. User no longer needs to compose on their own. Amanda will generate the right content based on the right context and the user’s answers to her questions.

Phase III  Total Control

At this point, Amanda will post things automatically for the user. The power of making their own decision is stripped away here from the user entirely.

Social Media

Amanda allows us to spend zero time on social media, but at the same time remain fully-active participants on social media. But this is done at the expense of us giving up our privacy, as Amanda requires constant video and audio monitoring of our surroundings in order to work. Ironically, the problem that Amanda sets out to solve would be more simply solved by refraining from social media use– not allow some artificially-intelligent agent to post for us.


The fact that Amanda fails to perfectly address our needs is, in fact, a criticism on our desire to embed technology in everything without first considering the limitations or associated dangers of that technology. AI is still in its infancy, yet we are so eager to find new applications for it. We’ve even gone so far as to develop AI for applications where human lives are at stake (see self-driving cars).


The tendency of Amanda to share more than we want to (snapping a picture and uploading it without asking for our approval) embodies the issues that come with the increasing flow of information throughout every aspect of our lives. We have become complacent with surrendering our privacy to welcome new technology, justifying it as a “necessary cost” associated with the increased convenience. That complacency is concerning– giving up our privacy isn’t a decision that we should take as lightly as absentmindedly clicking the “Allow” button.